The AI Slop Comes Before the Skill

A CEO’s perspective on why tool-learning velocity matters more than polished skepticism.

I think the workforce is splitting into two camps: people building fluency with AI, and people rejecting it on principle.

The pattern feels familiar.

During the Industrial Revolution, the Luddites resisted mechanized automation for reasons that were not entirely irrational: worker pay, loss of craft, deskilling, and lower-quality mass production. Some of those concerns rhyme with what we hear today about AI.

But when I see “AI slop,” I often see something different.

I see someone learning a new tool.

Engineers moved from slide rules to calculators. Writers moved from typewriters to computers. Teams moved from manual workflows to software systems. Now we are moving from software systems to AI-assisted systems.

Frontier models are tools.
Chatbots are tools.
Tasking agents are tools.
Integrating agents into workflows is a tool-use skill.

Most people are not going to be good at that skill immediately.

So when someone produces something that gets dismissed as “AI slop,” I try to judge it by a different standard: did AI help them produce a better outcome than they would have produced without it?

If a weak writer uses AI to produce something mediocre, but that mediocre output is better than what they could have written alone, that is not failure. That is progress.

That is a person putting in reps.

My own AI slop has been visual design.

I can build a decent systems engineering diagram. But when it comes to logos, artistic graphics, or visual metaphors, I am not naturally good at translating ideas into images. Historically, I had to pay an artist to turn my incoherent ramblings into something communicative.

Now, with generative image models, I can create imagery that is far better than what I could produce myself.

Is it obvious that some of my logos are AI-generated? Yes.
Are some of them AI slop? Probably.
Are they still more visually communicative than anything I could have made without AI? Absolutely.

I have been on sabbatical this year, and I have spent a lot of that time putting in my own AI reps. At first, that meant generating rough logos and forcing myself to publish many of the results so I would take each rep to completion.

It also meant getting called out by my brother when I used obviously AI-generated logos for otherwise deeply technical content.

Fair criticism. Necessary criticism.

But those reps are starting to compound. I am learning how to use AI to build the kind of artistic technical graphics that actually fit my style: visuals built from data, systems, structure, and engineering concepts.

That took months of bad reps before the output started getting better.

So when I see AI slop, I often see someone building fluency with a new tool that will pay dividends later in their career.

And when I see reflexive criticism of AI slop, I often see the opposite: inflexibility dressed up as taste.

As a CEO, I would rather hire the person demonstrating mental plasticity and tool-learning velocity than the person proudly refusing to adapt.

Not because all AI output is good.

Most first attempts are bad.

But the people putting in reps are going to get better.

The people sneering from the sidelines are not.